In endometrial and cervical cancer, MRI is primarily used for accurate staging, with T2-weighted imaging (T2WI) as a key sequence [1]. Over the past decade, diffusion-weighted imaging (DWI) has become widely accepted in routine clinical practice. Recently, radiomics has gained attention in cancer research, including gynecological cancers. In the context of endometrial cancer, models focusing on lymphovascular invasion (LVI), deep myometrial invasion, and lymph node metastasis have been major research subjects [2]. In cervical cancer, the main focuses have been on lymph node metastasis and treatment response to chemoradiation. Radiomics might prove most effective in addressing clinically relevant issues where current imaging diagnostics struggle, such as predicting LVI, lymph node metastasis, or treatment response.
Commonly cited drawbacks of radiomics include issues related to generalizability and reproducibility. While clinical implementation may take time, steady progress is being made in overcoming these challenges [2, 3]. In considering MRI radiomics in uterine cancer, vulnerability to artifacts is identified as a specific issue: motion artifacts from respiratory movements or intestines; susceptibility artifacts by rectal air often affecting DWI and its apparent diffusion coefficient (ADC) map. Motion artifacts by intestines is particularly noteworthy when performed without the administration of antispasmodics. Improving the robustness of T2WI and DWI to artifact is essential to ensure the generalizability of radiomics.
Several promising techniques have been reported for improving image quality. As a representative, deep learning (DL) reconstruction is gaining attention [4]. DL reconstruction enhances spatial resolution, reduces noise, and shortens acquisition time. The reduction of acquisition time, while maintaining image quality, directly leads to artifact resistance. Obtaining high-quality T2WI and DWI, even without antispasmodics, not only improves conventional imaging diagnostics but also benefits image analysis techniques, including radiomics. Furthermore, exploration of user-friendly clinical tools based on advanced computation techniques is crucial in future imaging diagnostics, extending beyond radiomics.

Acknowledgements

No acknowledgement found.